Abstract
648
Objectives: Brown Adipose Tissue (BAT) is a metabolically active adipose tissue that plays a major role in shivering-independent thermogenesis via uncoupling protein-1(UCP-1), as well as glucose and lipid homeostasis in adult humans.[1] Emerging radiomic analysis of tomographic imaging can further characterize BAT by providing a diverse set of novel quantitative features. These quantitative radiomic features have been utilized as biomarkers for phenotyping tumors, their response to treatment, and prognosis.[2] We aim to assess BAT radiomics features repeatability using FDG-PET in order to elucidate its radiomics characteristic
Methods: Twenty seven healthy adults (21 women and 6 men; mean age ± SD, 23.94 ± 3.64 y; body mass index, 22.21 ± 1.82 kg/m2) under the age of 35; 26 subjects completed the study protocol. Subjects were cooled via cool water cooling suit for 30 minutes prior to FDG administration and for an additional 60 minutes during FDG uptake phase. Each subject underwent FDG-PET/CT scan after cooling procedure. Participants underwent imaging with the same protocols and identical FDG dose within 2 weeks from the first scan. Brown Adipose Reporting Criteria in Imaging Studies (BARCIST) criteria was employed to identify BAT[3]. Active BAT in right supraclavicular region was selected. 878 radiomic features were extracted using Pyradiomics packages (http://pyradiomics.readthedocs.io) for each region of interest (ROI) on high definition reconstructed PET.[4] Lin's concordance correlation coefficients(CCC) was used to assess repeatability of the extracted radiomic features.[5] To elucidate the association between repeatable QIFs and SUVmax for the sake of characterization of the radiomic signature, a heat map analysis with hierarchical clustering on median centered radiomic features using complete linkage while using Pearson's correlation coefficient as the similarity measure.Results: 61 features(6.94%) had high reproducibility on test-retest (CCC>0.85). Hierarchical clustering of these repeatable features resulted in 4 distinct clusters. SUVmax did not cluster with any repeatable radiomic features.Conclusion:The identified FDG-PET radiomic cluster features provided additional information regarding BAT metabolic activation profile distinct from SUVmax and potentially can be used as quantitative biomarkers of BAT in future studies.
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.